W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2014

Re: Call for Consensus: Remove "reference set" from HPACK (to address #552)

From: Willy Tarreau <w@1wt.eu>
Date: Tue, 15 Jul 2014 11:52:34 +0200
To: Greg Wilkins <gregw@intalio.com>
Cc: Martin Thomson <martin.thomson@gmail.com>, Roberto Peon <grmocg@gmail.com>, HTTP Working Group <ietf-http-wg@w3.org>, Kazu Yamamoto <kazu@iij.ad.jp>, Tatsuhiro Tsujikawa <tatsuhiro.t@gmail.com>
Message-ID: <20140715095234.GD12333@1wt.eu>
Hi Greg,

On Tue, Jul 15, 2014 at 02:39:59PM +1000, Greg Wilkins wrote:
> I just cannot understand why you wish to create an entry in the dynamic
> table whenever the static table index is used for something like
> :status:200 ?    I only just accepted the reason that it limited the
> reference set size to 0 if the header table size was set to 0.  Now there
> is no reference set, there is no reason for this.

The benefit I found in doing so was to lower the static table indexes. In
fact, I'm seeing the dynamic table as a small cache and the static table
as the second level cache. Accessing recent data requires only a few bits,
accessing less recently used data (static table) requires a few more bits
sometimes leading to an extra byte.

I tend to find this design quite smart, because it manages to shrink the
working set representation simply by changing the indexes. I'm not sure
you'll be able to reach the same compression level without adding an
extra bit to the index number. That's a pure guess of course, but that's
what I find smart an efficient in this algorithm.

Received on Tuesday, 15 July 2014 09:53:06 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 March 2016 09:57:09 UTC